RESEARCH Open Access

Comparison of two ASC-derivedtherapeutics in an in vitro OA model:secretome versus extracellular vesiclesChiara Giannasi1*† , Stefania Niada1†, Cinzia Magagnotti2, Enrico Ragni3, Annapaola Andolfo2 andAnna Teresa Brini1,4

Abstract

Background: In the last years, several clinical trials have proved the safety and efficacy of adipose-derived stem/stromal cells (ASC) in contrasting osteoarthritis (OA). Since ASC act mainly through paracrine mechanisms, theirsecretome (conditioned medium, CM) represents a promising therapeutic alternative. ASC-CM is a complex cocktailof proteins, nucleic acids, and lipids released as soluble factors and/or conveyed into extracellular vesicles (EV). Here,we investigate its therapeutic potential in an in vitro model of OA.

Methods: Human articular chondrocytes (CH) were induced towards an OA phenotype by 10 ng/ml TNFα in thepresence of either ASC-CM or EV, both deriving from 5 × 105 cells, to evaluate the effect on hypertrophic, catabolic,and inflammatory markers.

Results: Given the same number of donor cells, our data reveal a higher therapeutic potential of ASC-CMcompared to EV alone that was confirmed by its enrichment in chondroprotective factors among which TIMP-1and -2 stand out. In details, only ASC-CM significantly decreased MMP activity (22% and 29% after 3 and 6 days)and PGE2 expression (up to 40% at day 6) boosted by the inflammatory cytokine. Conversely, both treatmentsdown-modulated of ~ 30% the hypertrophic marker COL10A1.

Conclusions: These biological and molecular evidences of ASC-CM beneficial action on CH with an induced OAphenotype may lay the basis for its future clinical translation as a cell-free therapeutic in the management of OA.

Keywords: Adipose-derived stem/stromal cells, Secretome, Extracellular vesicles, Chondrocytes, Osteoarthritis,Hypertrophy, MMP, PGE2

© The Author(s). 2020 Open Access This article is licensed under a Creative Commons Attribution 4.0 International License,which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you giveappropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate ifchanges were made. The images or other third party material in this article are included in the article's Creative Commonslicence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commonslicence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtainpermission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to thedata made available in this article, unless otherwise stated in a credit line to the data.

* Correspondence: chiara.giannasi@grupposandonato.it†Chiara Giannasi and Stefania Niada contributed equally to this work.1Laboratorio di Applicazioni Biotecnologiche, IRCCS Istituto OrtopedicoGaleazzi, Milan, ItalyFull list of author information is available at the end of the article

Giannasi et al. Stem Cell Research & Therapy (2020) 11:521 https://doi.org/10.1186/s13287-020-02035-5

BackgroundOsteoarthritis (OA) is an age-related disease that affectsmillions of people worldwide, representing a leadingcause of locomotor disability and therefore entailinghigh socio-economic costs [1]. Its pathogenesis is com-plex and engages different tissues. However, despite theproved involvement of subchondral bone and synoviumduring the degenerative process, the impairment of boththe structure and the function of articular cartilage isstill recognized as one of the earliest disease causingevents [2, 3].Currently, OA cannot be reversed pharmacologically,

but treatments can help relieve its symptoms. The medi-cations most commonly used in OA treatment are intra-articular corticosteroids, topical and oral non-steroidalanti-inflammatory drugs (NSAID), duloxetine, and acet-aminophen, often accompanied by physiotherapy and lifestyle modification [4, 5]. In the worst cases, when con-servative approaches fail, arthroplasty remains the onlyoption. Although the available treatments improve thequality of life in OA patients by reducing pain andpromoting joint mobility, the need to achieve adequatetissue regeneration and to develop drugs able to modifythe course of the disease (disease-modifying anti-OAdrugs, DMOAD) is still unmet. In this context, orthobio-logics are emerging as alternative therapeutic tools,thanks to their regenerative potential and cost-effectiveness [6, 7]. These approaches include intra-articular injection of platelet-rich plasma (PRP) andbiografts, such as autologous chondrocyte implantation(ACI), bone marrow concentrate (BMC), and adipose-derived stem/stromal cell (ASC) therapy. The lattertechniques rely on the presence, in both bone marrowand fat, of progenitor cells called mesenchymal stem/stromal cells (MSC). In response to various stimuli,MSC can differentiate into specialized cell types and/orbehave as “signaling” cells, able to pour into the micro-environment several mediators, such as nucleic acids,proteins, and lipids, that orchestrate the regenerativeprocess by modulating the immune system and recruit-ing specialized effectors (e.g., mast cells and T lympho-cytes). In the last years, in vitro [8–10] and in vivo [11–13] studies have proved the therapeutic potential ofMSC in counteracting cartilage damage and, to date,more than 100 clinical trials have evaluated/are assessingthe safety and efficacy of MSC intra-articular injectionin OA patients (http://www.clinicaltrials.gov). Sincenowadays it is widely accepted that MSC action is largelymediated by paracrine mechanisms [14], the scientificinterest has shifted towards the study of their secretome,the conditioned medium (MSC-CM). Indeed, cell secre-tome is a cocktail of soluble factors and extracellularvesicles (EV) with a promising potential in regenerativeapplications. EV are particles naturally released from the

cell that are delimited by a lipid bilayer and may be ofboth endosomal origin or plasma membrane-derived[15]. Since a consensus has not emerged on specificmarkers of EV subtypes yet, the recent nomenclatureestablished by the International Society for ExtracellularVesicles (ISEV) divides EV into small (< 200 nm) andlarge (> 200 nm) particles, previously called exosomesand microvesicles based on their endosomal or plasmamembrane origin [16]. Two works have recentlyreviewed the available evidence of MSC-CM therapeuticaction on cartilage, subchondral bone and synovium [17,18]. Among other MSC sources, adipose tissue presentsseveral advantages in terms of harvesting procedure, cellisolation, and expansion [19]. ASC efficacy and safetyhave been largely studied, both in vitro and in vivo, andconfirmed by clinical trials [20, 21]. Moreover, ASCtherapy in the treatment of COVID-19 disease has re-cently shown promising outcomes [22, 23]. In recentyears, our group investigated and characterized ASC-CM content in terms of both soluble factors [24] andvesicular components [25, 26]. Furthermore, we evalu-ated ASC-CM effects in vitro on a model of human ar-ticular chondrocytes (CH) induced towards an OA-likephenotype by the inflammatory cytokine TNFα [27]. Inour previous study, we proved that ASC-CM containshigh levels of chondroprotective factors and exertsshort-term anti-hypertrophic and anti-catabolic effectson TNFα-treated CH, confirming the potential of thiscell-free approach in the management of OA. Thepresent work aims at disclosing which components ofASC secretome play the major role in its beneficial ac-tion, by comparing the effects of ASC-CM and ASC-EVderiving from 5 × 105 cells in the same OA in vitromodel.

MethodsUnless otherwise stated, reagents were purchased fromSigma-Aldrich, St. Louis, MO, USA.

Cell culturesCell cultures were obtained from waste tissues collectedat IRCCS Istituto Ortopedico Galeazzi upon InstitutionalReview Board approval. Written informed consent wasobtained from all donors. In detail, ASC (1 male and 3females; 43 ± 15 years old) and CH (4 males and 3 fe-males, 64 ± 13 years old) were isolated from patientsundergoing esthetic or prosthetic surgery, followingwell-established protocols [24, 27, 28]. Briefly, aftermechanical fragmentation of the subcutaneous adiposetissue deriving from abdominoplasty surgery (n = 3) orabdominal liposuction (n = 1), ASC were isolated by en-zymatic digestion with 0.75 mg/ml type I Collagenase(Worthington Biochemical Corporation, Lakewood, NJ,USA) for 30 min and filtering of the stromal vascular

Giannasi et al. Stem Cell Research & Therapy (2020) 11:521 Page 2 of 15

fraction through a 100-μm cell strainer (Corning Incor-porated, Corning, NY, USA) [29]. All ASC donors werenormal-weight subjects (BMI < 30, no documented diag-nosis of obesity). CH derived from the femoral head ofOA patients who underwent total hip replacement: onlythe areas of macroscopically healthy cartilage (white,shiny, elastic, and firm) were harvested through a scalpeland digested overnight at 37 °C with 1.5 mg/ml type IICollagenase (Worthington Biochemical Corporation,Lakewood, NJ, USA) [28, 30]. The areas characterized byirregular surface, discoloration or softening were nevercollected, even at the cost of losing the entire sample, inorder to exclude any experimental bias linked to the useof strongly compromised cartilage. Cells were culturedin high glucose DMEM supplemented with 10% FBS(Euroclone, Pero, Italy), 2 mM L-glutamine, 50 U/mlpenicillin, and 50 μg/ml streptomycin at 37 °C in a hu-midified atmosphere with 5% CO2. The culture mediumwas further implemented with 110 μg/ml sodium pyru-vate for CH maintenance. Prior serum starvation for CMand EV production, ASC were characterized as previ-ously described [31–34] and their features are summa-rized in Supplementary Table 1.

CM and EV productionConditioned medium was collected from ~ 90% conflu-ent ASC (IV to VI passage) cultured for 72 h understarving conditions (absence of FBS), following opti-mized procedures [24]. Cells were monitored every day.No signs of cell suffering (e.g., detaching) was ever ob-served and cell viability was maintained for the wholestarving duration (data not shown), consistently with arecent report by Petrenko et al. [35] After 72 h, condi-tioned media were collected and centrifuged at 2500×gfor 15 min at 4 °C to remove dead cells, large apoptoticbodies, and debris. The supernatants were split in half toobtain coupled CM and EV samples, while donor cellswere counted in order to correlate cell number to theappropriate treatment volumes. An aliquot of condi-tioned medium was centrifuged for 90 min at 4000×g,4 °C, inside Amicon Ultra-15 Centrifugal Filter Deviceswith 3-kDa cut-off (Merck Millipore, Burlington, MA,USA), resulting in a 40–50-fold more concentrated finalproduct. This procedure leads to a final product whosesafety and efficacy have been already shown bothin vitro [27] and in vivo [36]. In parallel, EV isolationwas performed starting from naïve conditioned mediumthrough differential centrifugation at 100,000×g, 4 °C[25]. Both final products were characterized as follows.

Secretome characterizationNanoparticle tracking analysis (NTA)Coupled ASC-CM and ASC-EV samples were appropri-ately diluted in 0.22 μm triple-filtered PBS and analyzed

by NanoSight NS300 (Malvern PANalytical, Salisbury,UK). For each measurement, 3 videos lasting 1 min werecaptured. All measurements matched the quality criteriaof 20–120 particles/frame, concentration of 106–4 × 109

particles/ml and valid tracks > 20%. Upon capture, vid-eos were analyzed by the in-build NanoSight SoftwareNTA.

CytofluorimetryPrior to cytofluorimetry analysis, ASC-CM and ASC-EVwere appropriately diluted in 0.22 μm triple-filtered PBSand stained with the green fluorescent dye CFSE (carbo-xyfluorescein diacetate succinimidyl ester). In details,ASC-derived products were incubated with 20 nM CFSEfor 1 h at 37 °C [37], then analyzed without any furtherwashing. All data were obtained using a CytoFLEX flowcytometer (Beckman Coulter, Brea, CA, USA). At first,instrument calibration was set using Megamix-Plus SSC(Biocytex, Marseille, France), a reference bead mixturecomposed of FITC fluorescent spheres of heterogeneousdimensions (160 nm, 200 nm, 240 nm, and 500 nm).CFSE-positive (CFSE+) ASC-CM was run to set in SSC-H and FITC-H channels the region where to expect ourevents, accordingly to the coordinates given by thestandardization beads (Fig. 1b). CFSE+ samples werethen incubated for 20 min at 4 °C in the dark with APC-conjugated antibodies raised against CD9, CD63, andCD81 (BioLegend, San Diego, CA, USA, dilution 1:20)EV markers, as per well-documented protocols and gen-eral ISEV guidelines for positive EV characterization [16,38], then 1:2 diluted in 0.22 μm triple-filtered PBS andacquired for 300 s at a low flow rate [39]. PBS-dilutedantibodies and unlabeled samples were used as appropri-ate controls.

Transmission electron microscopyPBS-resuspended EV were absorbed for 10 min on For-mvar carbon-coated grids, and excess liquid was re-moved by a filter paper. Two percent uranyl acetatesolution was used as negative stain for 10 min and excessof liquid was removed by a filter paper. The grid wasdried at room temperature. Eventually, absorbed EVwere examined with a TALOS L120C transmission elec-tron microscope (Thermo Fisher Scientific, Waltham,MA, USA) at 120 kV.

Western blotting of secretome samplesFor CM and EV samples, protein concentration was de-termined using Bio-Rad Protein Assay (Bio-Rad, Milan,Italy) following standard procedures. ASC-CM and -EVsamples presented a total protein concentration of0.49 ± 0.27 and 0.11 ± 0.04 μg/μl, respectively. For West-ern Blot analyses, specimens were lysed in Laemli bufferand analyzed as exhaustively described in [25, 26].

Giannasi et al. Stem Cell Research & Therapy (2020) 11:521 Page 3 of 15

Briefly, proteins from CM and EV lysates deriving from106ASC (respectively 34.1 and 1.4 μg for ASC-CM and-EV) were resolved into 12% SDS-PAGE, transferredonto a nitrocellulose membrane, stained for 1 min withPonceau S, and then probed for the expression of thetypical EV markers HSP70 (ExoAb, System Biosciences,

Palo Alto, CA, USA, dilution 1:1000), FLOT1 (BD Bio-sciences, San Jose, CA, 250 μg/ml, 1:500 diluted), andCD9 (ExoAb, System Biosciences, Palo Alto, CA, USA,dilution 1:1000) [16, 40]. Specific signals were revealedafter incubation with appropriate secondary antibodies(Mouse IgG Secondary Antibody, Thermo Fisher

Fig. 1 Characterization of ASC-CM and -EV. a Representative images of NTA referred to ASC-CM (left) and ASC-EV (right). The table shows thedimensional parameters of the samples expressed as mean ± SD of 6 NTA measurements. b Flow cytometer calibration with standard beads andCFSE+ ASC-CM. The FITC+ gate encloses the coordinates in SSC-H and FITC-H channels where to expect the events of interest. c–e CD63, CD81,and CD9 staining of representative CFSE+ ASC-CM and ASC-EV samples. f Transmission electron microscopy image showing the characteristicmorphology of EV. The scale bar indicates 200 nm. g Representative Western blot of ASC-CM and EV lysates deriving from 106 ASC. Cell lysatefrom 5 × 104 ASC is shown as control. h Laser scanning confocal microscopy of CH treated with ASCGFP+-EV for 3 days. β-Tubulin was revealedwith an Alexa Fluor® 568 conjugated antibody (red), nuclei were stained with DAPI (blue) (magnification × 63). The scale bar indicates 10 μm andthe orthogonal views were obtained by Fiji software. i Total protein content per million ASC (μg/106 cells). Data are shown as mean ± SD (n = 4).l Ponceau S staining of ASC-CM and -EV lysates from 106 ASC. Cell lysate from 5 × 104 ASC is also shown

Giannasi et al. Stem Cell Research & Therapy (2020) 11:521 Page 4 of 15

Scientific, Waltham, MA, USA, 1.5 μg/μl, 1:2000 diluted,or ExoAb Rabbit Secondary Antibody System Biosci-ences, Palo Alto, CA, USA, dilution 1:20,000), followedby detection with ECL Westar Supernova (Cyanagen,Bologna, Italy). Images were acquired with ChemiDocImaging System (Bio-Rad, Milan, Italy).

Confocal laser scanning microscopyAs a proof of concept of EV internalization in our sys-tem, EV deriving from ASCGFP+ [41] (kindly provided byDr. Giulio Alessandri of IRCCS Neurological InstituteCarlo Besta, Milan) were isolated through standard pro-cedures and administered in vitro to CH seeded on glasscoverslips. After 3 days, specimens were fixed in 4%paraformaldehyde, permeabilized with 0.1% Triton X-100, and incubated overnight at 4 °C with a monoclonalantibody raised against β-Tubulin (Sigma-Aldrich, St.Louis, MO, USA, 2mg/ml, 1:100 diluted). The next day,samples were incubated at room temperature for 45 minwith a goat anti-mouse secondary antibody conjugatedwith Alexa Fluor® 568 (Abcam, Cambridge, UK, 2 mg/ml, 1:1000 diluted). After 3 washes, coverslips were thenmounted using ProLong™ Diamond AntifadeMountantwith DAPI (Thermo Fisher Scientific, Waltham, MA,USA) and analyzed by the confocal laser scanningmicroscope TCS SP8 (Leica Microsystems CMS GmbH,Wetzlar, Germany) using a × 63 objective. The obtainedimages were processed with Las X (Leica MicrosystemsCMS GmbH, Wetzlar, Germany) and analyzed with Fijisoftware.

In vitro OA induction and treatmentsIn our experimental set up, CH were always employed at1st culture passage in order to prevent their de-differentiation [42]. We clarified this aspect in the textfollowing the reviewer’s suggestion. Briefly, CH wereseeded at the density of 104 cells/cm2 in tissue culture-treated 6-well plates (Corning Incorporated, Corning,NY, USA) and cultured in complete medium until thefull confluence was reached [43], then shifted in acomplete medium containing 1% FBS, treated with 10ng/ml TNFα to mimic OA microenvironment [27, 44]and CM (38.7 ± 16 μl) or EV (6.8 ± 2.2 μl) from 5 × 105

ASC. OA was induced by TNFα for 3 and 6 days, con-currently with CM or EV treatment. At day 3 or 6, with-out any media change, supernatants were collected andcells lysed for further analyses.

Western blotting of CH samplesCH were lysed in 50 mM Tris-HCl (pH 7.5), 150 mMNaCl, 1% NP-40, and 0.1% SDS supplemented with pro-tease inhibitor cocktail (PIC) and 2mM PMSF. Upon in-cubation on ice for 30 min, lysates were centrifuged for15 min at 15,000×g, 4 °C, in order to eliminate cell

membranes and collect the cytosolic fraction. The pro-tein content of each sample was quantified through BCAAssay (Thermo Fisher Scientific, Waltham, MA, USA).Measurements were performed in technical duplicates.Samples were analyzed by 10% SDS-PAGE and Westernblotting (WB), using standard protocols [27]. For eachsample, 10 μg of protein extract were loaded and probedwith the following primary antibodies: rabbit anti-Collagen X (Thermo Fisher Scientific, Waltham, MA,USA, dilution 1:100), mouse anti-MMP13 (ThermoFisher Scientific, Waltham, MA, USA, 0.4 μg/μl, 1:100diluted), rabbit anti-MMP3 (Cell Signaling, Danvers,MA, USA, dilution 1:1000), rabbit anti-Connexin 43(Cell Signaling, Danvers, MA, USA, dilution 1:1000), andgoat anti-GAPDH (Santa Cruz Biotechnology, 0.1 μg/μl,1:1000 diluted). Specific bands were revealed upon incu-bation with appropriate secondary antibodies conjugatedto horseradish peroxidase (Rabbit IgG Secondary anti-body, Thermo Fisher Scientific, Waltham, MA, USA,dilution 1:10,000; Mouse IgG Secondary Antibody,Thermo Fisher Scientific, Waltham, MA, USA, dilution1:6000; Goat IgG Secondary Antibody, Santa Cruz Bio-technology, CA, USA; 0.1 μg/μl, 1:6000 diluted) followedby detection with ECL Westar Supernova (Cyanagen,Bologna, Italy). After image acquisition with ChemiDocImaging System, protein expression was quantifiedthrough Image Lab Software (Bio-Rad, Milan, Italy). Tonormalize target protein expression, the band intensityof each sample was divided by the intensity of the load-ing control protein GAPDH. Then, the fold change wascalculated by dividing the normalized expression fromeach lane by the normalized expression of the controlsample (CTRL = 1).

Analyses of culture supernatantsCH culture supernatants were collected and centrifugedfor 5 min at 2000×g, 4 °C, to remove dead cells and deb-ris, aliquoted and stored at − 20 °C. MMP activity wasassessed with SensoLyte 520 Generic MMP Activity Kit(AnaSpec, Fremont, CA, USA), following the standardprotocols. Briefly, pro-enzyme activation was performedthrough incubation with 1mM AMPA (4-aminophenyl-mercuric acetate) for 3 h at 37 °C in order to assesssimultaneously the activity of different MMP. Measure-ments were performed in technical duplicates. Thelength of this incubation was chosen according to themanufacturer’s instructions as the preferential activationtime to assess the activity of MMP-1 and -3, bothstrongly involved in OA. Samples were then incubatedwith the appropriate substrate for 45 min to run the en-zymatic reaction and the resulting fluorescence signal(excitation λ = 490 nm, emission λ = 520 nm) was readwith Wallac Victor II (Perkin Elmer, Milan, Italy).ADAMTS4 activity was tested using SensoLyte 520

Giannasi et al. Stem Cell Research & Therapy (2020) 11:521 Page 5 of 15

Aggrecanase-1 Assay Kit (AnaSpec, Fremont, CA, USA),following standard procedures. PGE2 levels wereassessed through Prostaglandin E2 Human CompetitiveELISA Kit (Thermo Fisher Scientific, Waltham, MA,USA) following the kit instructions; then, data were ana-lyzed with MyAssays analysis tool (https://www.myas-says.com). Measurements were performed in technicalduplicates.

nLC-MS/MS of ASC-CM and -EVASC-CM and ASC-EV samples were analyzed by differ-ential proteomics. Twenty micrograms of total proteinsfrom each sample were in-solution digested using filter-aided sample preparation (FASP) protocol, as reportedin literature [45]. Aliquots of the samples containingtryptic peptides were desalted using StageTip C18(Thermo Fisher Scientific, Bremen, Germany) and ana-lyzed by nLC-MS/MS using a Q-Exactive mass spec-trometer (Thermo Fisher Scientific, Bremen, Germany)equipped with a nano-electrospray ion source (ProxeonBiosystems, Odense, Denmark) and a nUPLC Easy nLC1000 (Proxeon Biosystems, Odense, Denmark). Peptideseparations occurred on a homemade (75 μm i.d., 15 cmlong) reverse phase silica capillary column, packed with1.9-μm ReproSil-Pur 120 C18-AQ (Dr. Maisch HPLCGmbH, Ammerbuch-Entringen, Germany). A gradientof eluents A (distilled water with 0.1% v/v formic acid)and B (acetonitrile with 0.1% v/v formic acid) was usedto achieve separation (300 nL/min flow rate). After 5min at 2% of B, the acetonitrile phase was increased upto 40% B in 83 min, followed by a wash step at 90% of B.Full scan spectra were acquired with the lock-mass op-tion, resolution set to 70,000 and mass range from m/z300 to 2000 Da. The ten most intense doubly and triplycharged ions were selected and fragmented. All MS/MSsamples were analyzed using Mascot (version 2.6, MatrixScience) search engine to search the human_proteome20190703 (96,470 sequences; 38,319,731 residues).Searches were performed with the following settings:trypsin as proteolytic enzyme, 2 missed cleavagesallowed, carbamidomethylation on cysteine as fixedmodification, protein N-terminus-acetylation and me-thionine oxidation as variable modifications, and masstolerance was set to 5 ppm and to 0.02 Da for precursorand fragment ions, respectively. To quantify proteins,the raw data were loaded into the MaxQuant [46] soft-ware version 1.6.1.0. Label-free protein quantificationwas based on the intensities of precursors. The experi-ments were performed in technical triplicates. Data areexpressed as label-free quantification (LFQ) intensity,count per second (cps). In order to identify differencesbetween ASC-CM and -EV that can be relevant in theOA context, the list of proteins quantified by nLC-MS/MS was run using the following keywords: Chondro-,

Metabol-, Catabol-, Inflamm-, and Matrix. The keywordswere chosen considering that OA is an inflammatorydisease affecting the osteochondral unit and alteringextracellular matrix metabolism and catabolism. Thefunctional enriched processes were then identified usingSTRING (Search Tool for the Retrieval of InteractingGenes/Proteins) (https://string-db.org/). Each identifiedprocess is reported in Supplementary Table 2 along withits GO identifier, the number of mapped genes withinour dataset, the number of mapped genes in the refer-ence dataset, its p value, and the list of gene nameswithin our dataset. Using the gene names assigned toeach of the five selected functional processes and theirprotein abundance levels measured by nLC-MS/MS,principal component analysis (PCA), and heat mapswere obtained using XLSTAT software. To validatenLC-MS/MS data, the presence of selected molecules inASC secretome was confirmed by immunoassaysthrough the Bio-Plex Multiplex System (Bio-Rad, Milan,Italy). In details, OPG and DKK-1 were quantified withthe Human Bone Magnetic Bead Panel-Bone Metabol-ism Multiplex Assay (HBNMAG-51K, Millipore, Bur-lington, MA, USA), MMP1 and 2 with the HumanMMP Magnetic Bead Panel 2 (HMMP2MAG-55 K,Millipore, Burlington, MA, USA), while TIMP-1,- 2 and-3 with the Human TIMP Magnetic Luminex Perform-ance Assay (LKTM003, R&D Systems, Minneapolis,MN, USA). Technical duplicates were analyzed for eachCM sample following previously described procedures[27, 47] and data analysis was performed with theMAGPIX xPONENT 4.2 software (Luminex Corpor-ation, Austin, TX, USA). The concentration of the se-lected molecules is reported in Supplementary Table 3along with the processes in which each factor is in-volved, as indicated in Supplementary Table 2.

StatisticsStatistical analysis was performed by one-way analysis ofvariance (ANOVA) using Tukey’s post hoc test in caseof normally distributed measures, otherwise (i.e., PGE2and COL10A1 data) by Friedman’s test followed byDunn’s multiple comparison. Differences were consid-ered significant at p ≤ 0.05. Unless otherwise stated, dataare expressed as mean ± SD of 5–7 independent experi-ments. All the analyses were performed using Prism 5(GraphPad Software, La Jolla, CA, USA).

ResultsASC-CM and -EV characterizationNTA demonstrates a comparable size distribution be-tween ASC-CM and ASC-EV samples (Fig. 1a), with the50% of events falling inside the dimensional range of147 ± 14 and 133 ± 19 nm, respectively. Given the samenumber of donor ASC, the two preparations differ for

Giannasi et al. Stem Cell Research & Therapy (2020) 11:521 Page 6 of 15

particle concentration, with a post-ultracentrifugationrecovery of about 30% of the input (ASC-CM = 9.0 ±4.3 × 108 particles/106cells and ASC-EV = 2.6 ± 0.9 × 108

particles/106cells). Flow cytometry confirms a similar EVsize (Fig. 1b) and a comparable expression of vesicularmarkers (Fig. 1c–e). In details, most of the vesicles areincluded within 240 nm (Fig. 1b) and the percentage ofpositive events for the EV markers CD63 (> 79%), CD81(> 76%), and CD9 (> 28%) is alike in ASC-CM and -EV(Fig. 1c–e). Transmission electron microscopy supportedthe presence of nanoparticles with the characteristiccup-shaped morphology and the expected size range(Fig. 1f). Furthermore, protein expression of HSP70,Flotillin-1 (FLOT1), and CD9 shows a similar vesicularphenotype (Fig. 1g). ASC-EV were able to interact withrecipient cells: indeed in Fig. 1h, as a proof of concept,the incorporation of EV derived by ASCGFP+ [41] in CHis shown. The spatial co-localization of the green (EV orEV cluster from ASCGFP+) and the red (cytoskeleton)signals shown in the orthogonal views suggests the in-corporation of vesicular elements in the recipient cell.Additional evidence is provided in Supplementary Figure1. At last, ASC-CM contains 25-fold more proteins (interms of quantity) than EV samples, indicating that sol-uble factors are also abundant (Fig. 1i, l).

ASC-CM, but not ASC-EV, significantly reduces TNFα-induced MMP activityUnstimulated CH secrete low levels of active matrix me-talloproteinases (MMP), but their activity is strongly in-creased by the inflammatory stimulus at both timepoints (+ 2498% and + 1781%, respectively) (Fig. 2a).ASC-CM significantly reverts TNFα-induced activationof about 22% and 29% at day 3 and 6 (Fig. 2a, left andright panel, respectively). In contrast, no effect wasexerted by the treatment with EV. Differential proteo-mics allowed the quantification of TIMP-1 and -2, thetwo most abundant TIMP (tissue inhibitors of MMP) inASC secretome [24, 27] (Fig. 2b). These data, obtainedanalyzing the same amount of proteins for ASC-CM and-EV, revealed that both inhibitors are slightly more rep-resented in CM samples (Fig. 2b). Actually, CM containsfar more TIMP than EV per ASC number, since EVsamples are derived from a 25-fold higher number ofcells. The fact that CM acts mainly through the presenceof MMP inhibitors is further supported by the lack of ef-fect on MMP expression. Protein expression of MMP-13and MMP-3, two matrix-degrading enzymes involved inOA [48–50], is displayed in Fig. 2c and d and in Supple-mentary Figure 2. Despite the large inter-donor variabil-ity due to the use of patient-derived articularchondrocytes, a clear effect of TNFα on MMP expres-sion is always present. Differently, CM exerts no effecton their overexpression, as previously shown [27], nor

did EV. ADAMTS-4 activity was also tested, since TIMPact also on aggrecanases. However, in our experimentalsetting, its activity was always undetectable (data notshown).

ASC-CM and ASC-EV differently modulate inflammationand hypertrophy markers in TNFα-treated CHIn order to investigate CH activation by TNFα and aneffect of ASC secretome, we examined PGE2 release,total protein content, Collagen X, and Connexin 43expression. As expected, TNFα raises the extracellularconcentration of the inflammatory mediator PGE2, scal-ing it more than 2 orders of magnitude at both timepoints (Fig. 3a). Since high PGE2 concentrations can in-hibit proteoglycan synthesis and stimulate matrix deg-radation [51, 52], a possible counteracting effect of ASCsecretome was hypothesized. ASC-CM decreased PGE2upregulation up to 40% at day 6 (Fig. 3a, right panel).Conversely, ASC-EV did not affect TNFα-induced PGE2levels.By contrast, both ASC-derived treatments partly blunt

TNFα effect (about − 30%) on the production of Colla-gen type X (COL10A1, Fig. 3b), a short chain collagenexpressed by hypertrophic CH [50]. EV induced a morelong-lasting effect while ASC-CM acted incisively onlyat the early time point (Fig. 3b, right and left panel, re-spectively, Supplementary Figure 2).The TNFα-induced phenotypic shift towards hyper-

trophy can be inferred by the significant increase in CHtotal protein content (Fig. 3c, + 38%), due to an increasein cell proliferation. ASC-CM did not counteract TNFα-induced cell growth, while, at day 6, ASC-EV act in syn-ergy with the inflammatory cytokine, fostering its pro-proliferative action (+ 55% vs CTRL, + 12% vs TNFαalone) (Fig. 3c, right panel).At last, the expression of Connexin 43 (Cx43), the

most widely expressed connexin in the musculoskeletalsystem [53, 54], was investigated. Its levels are clearlydown-modulated by TNFα (Fig. 3d, Supplementary Fig-ure 2), especially in the long run (− 60% vs CTRL, Fig. 3dright panel). ASC-EV further reduced Cx43 expressionat day 3 (− 16% vs TNFα alone, Fig. 3d left panel) whilethe effect of ASC-CM is negligible.

ASC-CM and ASC-EV present different factors of interestin the OA contextProteomic data analysis through OA-related keywordsconfirms that CM and EV protein profiles are distinctfor all the considered processes (Fig. 4 and Supplemen-tary Table 4). Of note, the analysis on chondroitin sul-fate factors led to an important discrimination betweenCM and EV, with more than 81.9% of variance explainedby factors 1 and 2 (Fig. 4a, Supplementary Table 4). Themost relevant differences were distinguishable in the

Giannasi et al. Stem Cell Research & Therapy (2020) 11:521 Page 7 of 15

expression of Versican (greater than in ASC-EV) andDecorin and Biglycan (greater than in ASC-CM) (Fig. 4band Supplementary Table 5). At last, a manual check forother proteins linked to OA allowed the identification ofBone Morphogenetic Protein 1, Dickkopf-related protein3, and 4 members of the ADAM (A Disintegrin and Me-talloproteinase) family (in details ADAM10, ADAM12,ADAM17, and ADAM9) as more abundant in the CMsamples (Supplementary Table 5). PCA on the factorsassociated to catabolism, metabolism, matrix, and

inflammation allowed a distinction between CM and EV,with a score always higher than 60% of variance ex-plained by F1 and F2 (Fig. 4c–f and SupplementaryTable 4).

DiscussionWith the discovery that MSC engraftment and differen-tiation play a partial role in the success of cell therapy[55], over the years the scientific interest has shifted to-wards MSC-secreted factors. In the last decade, the

Fig. 2 Reduction of MMP activity by ASC-CM, TIMP quantification, and MMP expression. a MMP activity, analyzed in CH culture medium (n = 7independent experiments) 3 and 6 days after the treatments, is expressed as arbitrary fluorescence units (AFU). All conditions statistically differfrom control (at day 3: TNF p < .01, TNF+ASC-CM p < .05, and TNF+ASC-EV p < .001; at day 6: TNF p < .001, TNF+ASC-CM p < .05, and TNF+ASC-EVp < .001). Significance vs 10 ng/ml TNFα is shown as $p < .05; vs ASC-EV as #p < .05, ##p < .01. b TIMP-1 and 2 data are expressed as label-freequantification (LFQ) intensity, count per second (cps) from differential proteomic analysis of 20 μg of ASC-CM and -EV proteins. Means ± SD(n = 3) are shown. c, d Quantification of the expression of MMP-13 (c) and MMP-3 (d) in TNFα-stimulated and ASC-CM- or -EV-treated CH at day3 and 6 analyzed by Western blot. Data (n = 5 independent experiments) were normalized on GAPDH and expressed as relative values (CTRL = 1)

Giannasi et al. Stem Cell Research & Therapy (2020) 11:521 Page 8 of 15

number of studies focused on the physical and func-tional characterization of MSC secretome has grown ex-ponentially. Recently some cell-free products havereached the clinics in phase I and II trials [56] for di-verse applications in which the beneficial action of thecells of origin was already well-documented [20, 21, 57]:MSC-CM in wound healing [58], alopecia [59, 60], boneregeneration [61], and multiple sclerosis [62], whileMSC-EV in graft versus host disease [63], chronic kidney

disease [64], type 1 diabetes, macular holes, and acute is-chemic stroke (cited in [65]). Considering future clinicalapplications in the OA management, here, we comparedthe potential the potential of ASC whole secretome ver-sus its EV component. Our treatment strategy followswhat right now represents the gold standard for celltherapy, i.e., cell number-based dosage. This approachwas already reported in literature for EV administration,both in vitro [66] and in vivo [67]. At first, the

Fig. 3 Hypertrophy and inflammatory markers induced by TNFα treatment. a PGE2 levels, quantified in CH culture medium at day 3 and 6 aftertreatments, are expressed as ng/ml (n = 6 independent experiments). b COL10A1 expression by Western blot analysis. Data (n = 5 independentexperiments) were normalized on GAPDH and shown as relative values (CTRL = 1). Data in a and b were analyzed by Friedman’s test followed byDunn’s multiple comparison test and significance vs CTRL is shown as *p < .05. For each column, the box extends from the 25th to 75thpercentiles, the line in the middle is plotted at the median while the whiskers indicate minimum and maximum value. c CH proteinconcentration at day 3 and 6 (left and right panel, respectively). Significance vs CTRL is shown as *p < .05 and **p < .01; vs 10 ng/ml TNFα as$p < .05 (n = 7 independent experiments). d Cx43 expression by Western blot. Data (n = 5 independent experiments) were normalized on GAPDHand expressed as relative values (CTRL = 1). Significance vs CTRL is shown as *p < .05 and **p < .01

Giannasi et al. Stem Cell Research & Therapy (2020) 11:521 Page 9 of 15

assumption that ASC-CM preparation allows a completeretention of the vesicular components [27, 68] has beenvalidated. Then, we compared the two preparations fo-cusing on size distribution, the presence of EV markers[16, 40] and particle concentration. Our data confirmthat the EV isolation procedure through ultracentrifuga-tion neither affected the quality of the particles norenriched any subpopulation. Indeed, we confirmed asimilar vesicular profile between ASC-CM and -EV interms of dimensions, antigen expression, and granular-ity/complexity. Moreover, we give evidence of a 3 timeshigher vesicular yield in concentrated CM samplescompared to EV ones, due to particle loss during the

ultracentrifugation procedure [69–71]. Therefore, ourin vitro data compare two preparations deriving fromthe same number of donor cells, with CM accountingfor both soluble factors and a higher number of retainedparticles in comparison to ultracentrifuge-isolated EV.Conversely, our differential proteomic analysis considersthe same protein amount for ASC-CM and -EV (20 μg/sample).Both in vitro evidence and the differential analysis of

the protein content between the two preparations sug-gest a higher therapeutic anti-OA potential of ASC-CMover ASC-EV. One of the most relevant differences wasthe lack of inhibition of MMP activity by ASC-EV,

Fig. 4 PCA and heat map of chondroitin sulfate-related factors. PCA plots of the samples based on 10 chondroitin sulfate- (a), 56 inflammation-(c), 425 catabolism- (d), 459 metabolism- (e), and 169 matrix- (f) mapped gene names. Color scale: red (down-represented) to green (up-represented) through black. (b) Heat map for the chondroitin sulfate process-associated genes, constructed on the basis of protein abundancelevels estimated by nLC-MS/MS in 3 ASC-CM (ASC-CM 1-3) and 3 ASC-EV (ASC-EV 1-3) samples. Principal component analysis (PCA) and heatmaps were obtained using XLSTAT software. F1 and F2, factor 1 and 2

Giannasi et al. Stem Cell Research & Therapy (2020) 11:521 Page 10 of 15

confirming our previous assumption that the blunting ofMMP activity is a direct consequence of active TIMP inASC secretome [27]. Indeed, these inhibitors are morerepresented in the complete secretome compared to thevesicular fraction. To our knowledge, it is the first timethat the greater abundance of freely dissolved TIMP ra-ther than EV-released ones is clearly defined. This aspectgains relevance in the light of developing MMP inhibi-tors as potential pharmacological tools in the manage-ment of a variety of diseases [72]. It also points out thatfor every pathology implying the aberrant activation ofMMP [73, 74], the complete secretome, rather than theEV fraction alone, should be considered the more prom-ising therapeutic strategy.In addition, ASC-CM was more efficient in reducing

the release of the inflammatory mediator PGE2 byTNFα-stimulated CH. This prostaglandin is known toexert multiple opposed functions based on its concentra-tion. In our case, the down-modulation of PGE2 aimedat restoring its physiologic levels linked to a healthy CHphenotype. Indeed unstimulated chondrocytes releaselow amount of PGE2 (average release of about 20 pg/ml)that are consistent with the concentration known to in-hibit collagen cleavage and the expression of hyper-trophy markers [75]. By contrast, TNFα raised PGE2levels above the pro-anabolic concentration (average re-lease of about 2.5 ng/ml) and it is known that similarlevels (from 1 to 1000 ng/ml) exert a pro-catabolic andanti-anabolic effect on articular chondrocytes [52, 76].As follows, the reduction induced by ASC-CM appearsbeneficial. Vonk et al. [77] observed a more markedeffect of CM compared to EV in reducing TNFα down-stream effectors by CH, in particular looking atcyclooxygenase-2 (COX2) expression. Indeed, TNFαtreatment induces PGE2 release through the activationof COX2 transcription via NF-κB [78]. However, ourdata (Supplementary Figure 3) show the lack of a clear-cut effect of ASC-CM on TNFα-induced COX2 proteinexpression at 3 and 6 days, suggesting that the mecha-nisms underlying the blunting of PGE2 may act at a dif-ferent level. Further investigations are currently ongoingto elucidate this aspect.In our opinion the observed minor ASC-EV effects

cannot be ascribed to the lack of internalization of EVinto recipient cells. Indeed, administering ASCGFP+-EVto stained CH, a clear intracellular co-localization of thefluorescent signals was observed, suggesting an efficientEV uptake. ASC-EV incorporation has been already re-ported in other in vitro systems [39, 66] and future in-vestigations will be necessary to disclose its underlyingmechanisms (e.g., endocytosis and interaction of cellsurface receptors [79]). Moreover, here, we show thatEV modulate hypertrophy markers, contrasting TNFαaction on COL10A1 expression to a similar—if not more

marked—extent as ASC-CM. EV effect on chondrocytesseems less donor-dependent and more long-lasting com-pared to ASC-CM, suggesting that the mediator ofCOL10A1 reduction may be stored in EV and its benefi-cial action can be reduced by soluble factors. The regula-tion of COL10A1 gene expression during CHhypertrophic differentiation depends on multiple factors,including both transactivators and repressors (such asRunx2 and Sox9, respectively) and has not been fullyelucidated yet [80]. However, in vitro evidences demon-strate that the overexpression of the miRNA hsa-miR-148a decreases COL10A1 levels together with two otherOA-related genes, MMP13 and ADAMTS5 [81]. Ofnote, the presence of hsa-miR-148a in EV derived fromboth naïve and IFNγ-primed ASC has recently beendemonstrated by Ragni et al., as shown in their Add-itional File 3 [82]. Moreover, here, we have identifiedVersican, a chondroitin sulfate proteoglycan, more abun-dant in EV. Interestingly, it plays important roles inchondrogenesis and in the retention of cartilage extra-cellular matrix (ECM) [83]. Even though EV reducedCOL10A1 expression, suggesting a potential modulationof hypertrophy, no reduction on TNFα-mediated in-creased metabolism and proliferation was depicted. Bycontrast, we observed a slight increase in cell growth,previously reported also by Vonk et al. [77]. However, intheir setting TNFα induced a reduction in chondrocyteproliferation that we have never observed in our in vitromodel [27]. We ascribed the increase of chondrocyteproliferation by TNFα to the reduction of Cx43 expres-sion, consistently with the fact that Cx43 C-terminal do-main (CTD) influences chondrocyte proliferation andphenotype maintenance [54]. Moreover, Cx43 reductionby TNFα has been shown in several cell types and plaus-ible causes could be the activation of ubiquitin-proteasome system [84] or the activation of JNK byTNFα [85]. Additionally, the cleavage of Cx43 CTD byMMP has been already documented [86] and further ex-periments will be performed with the aim of evaluatingwhether the reduced Cx43 signal depends either on themodulation of its protein expression or on a post-translational mechanism.Taken together, these results show that, given the

same number of donor cells, ASC-CM is more efficientthan ultracentrifuge-isolated EV. ASC-CM and ASC-EVpresent different protein composition not only in termsof quantity but also in terms of quality. Indeed, PCAshows specific differences in OA-related processes, span-ning from ECM maintenance to inflammation. In details,ASC-CM appears to be more abundant in cartilage pro-tective factors. Besides TIMP, we have identified otherproteins involved in ECM organization and chondrogen-esis. Among them, Biglycan (BGN) and Decorin (DCN),two small leucine-rich proteoglycans (SLRP), can bind

Giannasi et al. Stem Cell Research & Therapy (2020) 11:521 Page 11 of 15

different types of collagen and organize the fibrils. More-over, they interact with TGFβ1 and control its signaling[87]. Also, BMP-1 plays an important role in collagen fi-bril organization. It cleaves C-propeptides of procolla-gens I, II, and III allowing the correct incorporation ofmonomers into growing fibrils [88]. Furthermore, it acti-vates diverse TGFβ superfamily proteins that are funda-mental for ECM formation and growth factor activity[89]. DKK-3 exerts protective roles in OA cartilage byinteracting with TGFβ signaling, too [90]. Consideringthat, in the OA context, both collagen fibril organizationand TGFβ signaling are affected [91, 92], the soluble fac-tors present in ASC-CM could help in re-establishing aphysiological condition. ASC-CM is enriched also infour ADAM, a group of enzymes involved in jointhomeostasis. ADAM9 is involved in chondrogenesis[93], ADAM10 has a controversial functionality and itsexpression is increased in both developing and OA car-tilage. Since its substrate in the articular district has notbeen identified yet, its implications in pathological con-ditions are still unclear [94]. In addition, ADAM12seems to be associated to OA and chondrocyte matur-ation [94]. The presence in ASC-CM of TNFα-activatingpro-inflammatory ADAM17 could be dangerous; how-ever, both its inhibitors TIMP-3 and CD9 [95] arepresent in the secretome as well [27], suggesting thatthis enzyme could be inactivated. This considerationhighlights the complexity of secretome and the relevanceof the balance between all the different factors.

ConclusionsIn conclusion, our study suggests that soluble and EV-associated factors concur in a strongly synergistic man-ner to the anti-inflammatory, anti-hypertrophic, andanti-catabolic effect of ASC secretome. Even though wemainly investigated the differences in terms of proteincontent between the two cell products, we need to con-sider that they surely differ also in terms of nucleic acidssuch as miRNA and lipid composition. To date, theknowledge of the bioactive components eliciting a thera-peutic action in different preclinical scenarios is still verylimited, and their characterization in the perspective of aclinical translation is still ongoing. However, CM prepa-rations, compared to EV isolated by ultracentrifugation,present several advantages, such as a higher vesicularyield and a lower manipulation, accounting for an easiercompliance with good manufacturing practices (GMP)and a better scalability. Indeed, from a clinical point ofview, CM production is cheaper and faster than EV iso-lation by the currently available, routinely used, tech-niques (e.g., ultracentrifugation and size exclusionchromatography). Moreover, CM yields a wider array ofbioactive factors, soluble, freely dissolved proteins, nu-cleic acids, and lipids, with respect to the vesicular

fraction alone [66]. In the future, more complex in vitromodels, such as organoids or models considering theinteraction with other articular cell types (e.g., synovio-cytes and osteoblasts), should be set and additional pre-clinical investigations, focused on the optimization ofCM production, will be performed in order to confirmthe efficacy of ASC-CM in the OA context.

Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1186/s13287-020-02035-5.

Additional file 1: Supplementary Figure 1. Representative images ofEV incorporation by CH. EV derived from ASCGFP+ are indicated by greenarrows, β-Tubulin was revealed with an Alexa Fluor® 568 conjugated anti-body (red) and nuclei were stained with DAPI (blue). The scale bars indi-cate 10 μm and the orthogonal views referred to the EV encircled inyellow were obtained by Fiji software.

Additional file 2: Supplementary Figure 2. Representative WesternBlot membrane for COL10A1, MMP-13, MMP-3, Cx43 and GAPDH.

Additional file 3: Supplementary Figure 3. (A) Quantification of theexpression of COX2 (mAb #12282, Cell Signaling Technology, Danvers,MA, USA) in TNFα-stimulated and ASC-CM or -EV treated CH at day 3 and6 analyzed by Western Blot. Data (n = 3 independent experiments) werenormalized on GAPDH and expressed as relative values (CTRL = 1).(B) Rep-resentative Western Blot membrane for COX2 and GAPDH.

Additional file 4: Supplementary Table 1. ASC features. Details ofculture conditions (CTRL, Osteoinduction and Adipoinduction) andperformed assays are reported. Supplementary Table 2. Functionalenriched processes identified with STRING through OA-related keywords(Chondro-, Metabol-,Catabol-, Inflamm- and Matrix). GO: Gene ontologyidentifier. Supplementary Table 3.Quantification of OA-related factorsin ASC-CM (n = 3) expressed as mean ± SD. Details of the biological pro-cesses in which each factors is involved, according to SupplementaryTable 2, are also provided. Supplementary Table 4. Principal Compo-nent Analysis (PCA) details by XLSTAT software. Number of factors ana-lyzed for each OA-related keyword are reported in brackets.Supplementary Table 5. nLC-MS/MS quantification of OA-related fac-tors in ASC-CM and ASC-EV (n = 3). Factors significantly different betweenCM and EV samples are in bold. Proteins more abundant in EV that in CMare highlighted in gray.

AbbreviationsACI: Autologous chondrocyte implantation; ADAM: A disintegrin andmetalloproteinase; AMPA: 4-Aminophenylmercuric acetate; ASC: Adipose-derived stem/stromal cells; BGN: Biglycan; BMC: Bone marrow concentrate;BMI: Body mass index; CFSE: Carboxyfluorescein diacetate succinimidyl ester;CH: Articular chondrocytes; CM: Conditioned medium; COL10A1: Collagentype X; COX2: Cyclooxygenase-2; cps: Count per second; CTD: C-terminaldomain; Cx43: Connexin 43; DCN: Decorin; DMOAD: Disease-modifying anti-OA drugs; ECM: Extracellular matrix; EV: Extracellular vesicles; FASP: Filter-aided sample preparation; GMP: Good manufacturing practices; LFQ: Label-free quantification; MMP: Matrix metalloproteinases; MSC: Mesenchymalstem/stromal cells; NSAID: Non-steroidal anti-inflammatory drugs;NTA: Nanoparticle tracking analysis; OA: Osteoarthritis; PCA: Principalcomponent analysis; PIC: Protease inhibitor cocktail; PRP: Platelet-rich plasma;SLRP: Small leucine-rich proteoglycans; STRING: Search Tool for the Retrievalof Interacting Genes/Proteins; TIMP: Tissue inhibitors of matrixmetalloproteinases; WB: Western blotting

AcknowledgementsThe authors would like to thank Dr. Giulio Alessandri (IRCCS NeurologicalInstitute Carlo Besta, Milan) for providing the ASCGFP+ cell line and Dr.Alessandro Bidossi (IRCCS Istituto Ortopedico Galeazzi, Milan) for theassistance with confocal imaging and analysis.

Giannasi et al. Stem Cell Research & Therapy (2020) 11:521 Page 12 of 15

Authors’ contributionsC.G. and S.N. equally contributed to this work. Conceptualization, C.G., S.N.and A.T.B. Methodology, C.G., S.N., C.M., E.R., and A.A. Resources, C.G., S.N.,C.M., E.R., A.A., and A.T.B. Data curation, C.G. and S.N. Writing-original draftpreparation, C.G. and S.N. Writing-review and editing, C.M., E.R., A.A., andA.T.B. Supervision, A.T.B. Project administration, C.G., S.N., and A.T.B. Fundingacquisition, A.T.B. All authors have read and agreed to the published versionof the manuscript.

FundingThis research was funded by the Italian Ministry of Health, grant number RCL1027 and L1039, and by the University of Milan, Department of BiomedicalSurgical and Dental Sciences, grant number RV_RIC_AT16RWEIBN_02 andRV_LIB16ABRIN_M. Funds for open-access publication fees were receivedfrom Istituto Ortopedico Galeazzi, Milan, Italy.

Availability of data and materialsThe datasets used and/or analyzed during the current study are availablefrom the corresponding author on reasonable request.

Ethics approval and consent to participateUpon written informed consent and following the ethical principles of theHelsinki Declaration, human tissues were collected as waste material fromthe surgery room performed at IRCCS Istituto Ortopedico Galeazzi underInstitutional Review Board approval (number 6/int/2018 approved by SanRaffaele Hospital Ethics Committee).

Consent for publicationNot applicable.

Competing interestsThe authors declare that they have no competing interests.

Author details1Laboratorio di Applicazioni Biotecnologiche, IRCCS Istituto OrtopedicoGaleazzi, Milan, Italy. 2Proteomics and Metabolomics Facility (ProMeFa), IRCCSSan Raffaele Scientific Institute, Milan, Italy. 3Laboratorio di BiotecnologieApplicate all’Ortopedia, IRCCS Istituto Ortopedico Galeazzi, Milan, Italy.4Department of Biomedical, Surgical and Dental Sciences, Università degliStudi di Milano, Milan, Italy.

Received: 9 September 2020 Accepted: 18 November 2020

References1. Salmon JH, Rat AC, Sellam J, Michel M, Eschard JP, Guillemin F, et al.

Economic impact of lower-limb osteoarthritis worldwide: a systematicreview of cost-of-illness studies. Osteoarthr Cartil. 2016;24(9):1500–8.

2. Goldring SR, Goldring MB. Changes in the osteochondral unit duringosteoarthritis: structure, function and cartilage-bone crosstalk. Nat RevRheumatol. 2016;12(11):632–44.

3. Berenbaum F. Osteoarthritis as an inflammatory disease (osteoarthritis is notosteoarthrosis!). Osteoarthr Cartil. 2013;21(1):16–21.

4. McAlindon TE, Bannuru RR, Sullivan MC, Arden NK, Berenbaum F, Bierma-Zeinstra SM, et al. OARSI guidelines for the non-surgical management ofknee osteoarthritis. Osteoarthr Cartil. 2014;22(3):363–88.

5. Ragni E, Mangiavini L, Viganò M, Brini AT, Peretti GM, Banfi G, de GirolamoL. Management of Osteoarthritis During the COVID-19 Pandemic. ClinPharmacol Ther. 2020;108(4):719-29. https://doi.org/10.1002/cpt.1910.

6. Huebner K, Frank RM, Getgood A. Ortho-biologics for osteoarthritis. ClinSports Med. 2019;38(1):123–41.

7. Sherman BJ, Chahla J, Glowney J, Frank RM. The role of orthobiologics inthe management of osteoarthritis and focal cartilage defects. Orthopedics.2019;42(2):66–73.

8. Maumus M, Manferdini C, Toupet K, Peyrafitte JA, Ferreira R, Facchini A,et al. Adipose mesenchymal stem cells protect chondrocytes fromdegeneration associated with osteoarthritis. Stem Cell Res. 2013;11(2):834–44.

9. Manferdini C, Maumus M, Gabusi E, Piacentini A, Filardo G, Peyrafitte JA,et al. Adipose-derived mesenchymal stem cells exert antiinflammatory

effects on chondrocytes and synoviocytes from osteoarthritis patientsthrough prostaglandin E2. Arthritis Rheum. 2013;65(5):1271–81.

10. Manferdini C, Maumus M, Gabusi E, Paolella F, Grassi F, Jorgensen C, et al.Lack of anti-inflammatory and anti-catabolic effects on basal inflamedosteoarthritic chondrocytes or synoviocytes by adipose stem cell-conditioned medium. Osteoarthr Cartil. 2015;23(11):2045–57.

11. Lee KB, Hui JH, Song IC, Ardany L, Lee EH. Injectable mesenchymal stemcell therapy for large cartilage defects--a porcine model. Stem Cells. 2007;25(11):2964–71.

12. Xie X, Wang Y, Zhao C, Guo S, Liu S, Jia W, et al. Comparative evaluation ofMSCs from bone marrow and adipose tissue seeded in PRP-derived scaffoldfor cartilage regeneration. Biomaterials. 2012;33(29):7008–18.

13. Lim CT, Ren X, Afizah MH, Tarigan-Panjaitan S, Yang Z, Wu Y, Chian KS,Mikos AG, Hui JH. Repair of osteochondral defects with rehydrated freeze-dried oligo[poly(ethylene glycol) fumarate] hydrogels seeded with bonemarrow mesenchymal stem cells in a porcine model. Tissue Eng Part A.2013;19(15-16):1852-61. https://doi.org/10.1089/ten.TEA.2012.0621.

14. Fitzsimmons REB, Mazurek MS, Soos A, Simmons CA. Mesenchymal stromal/stem cells in regenerative medicine and tissue engineering. Stem Cells Int.2018;2018:8031718.

15. van Niel G, D'Angelo G, Raposo G. Shedding light on the cell biology ofextracellular vesicles. Nat Rev Mol Cell Biol. 2018;19(4):213–28.

16. Thery C, Witwer KW, Aikawa E, Alcaraz MJ, Anderson JD, AndriantsitohainaR, et al. Minimal information for studies of extracellular vesicles 2018 (MISEV2018): a position statement of the International Society for ExtracellularVesicles and update of the MISEV2014 guidelines. J Extracell Vesicles. 2018;7(1):1535750.

17. Mancuso P, Raman S, Glynn A, Barry F, Murphy JM. Mesenchymal stem celltherapy for osteoarthritis: the critical role of the cell secretome. FrontBioeng Biotechnol. 2019;7:9.

18. D'Arrigo D, Roffi A, Cucchiarini M, Moretti M, Candrian C, Filardo G.Secretome and Extracellular Vesicles as New Biological Therapies for KneeOsteoarthritis: A Systematic Review. J Clin Med. 2019;8(11):1867. https://doi.org/10.3390/jcm8111867.

19. Chu DT, Nguyen Thi Phuong T, Tien NLB, Tran DK, Minh LB, Thanh VV, GiaAnh P, Pham VH, Thi Nga V. Adipose Tissue Stem Cells for Therapy: AnUpdate on the Progress of Isolation, Culture, Storage, and ClinicalApplication. J Clin Med. 2019;8(7):917. https://doi.org/10.3390/jcm8070917.

20. Squillaro T, Peluso G, Galderisi U. Clinical trials with mesenchymal stem cells:an update. Cell Transplant. 2016;25(5):829–48.

21. Gentile P, Casella D, Palma E, Calabrese C. Engineered Fat Graft Enhancedwith Adipose-Derived Stromal Vascular Fraction Cells for RegenerativeMedicine: Clinical, Histological and Instrumental Evaluation in BreastReconstruction. J Clin Med. 2019;8(4):504. https://doi.org/10.3390/jcm8040504.

22. Sánchez-Guijo F, García-Arranz M, López-Parra M, Monedero P, Mata-Martínez C, Santos A, et al. Adipose-derived mesenchymal stromal cells forthe treatment of patients with severe SARS-CoV-2 pneumonia requiringmechanical ventilation. A proof of concept study. EClinicalMedicine. 2020;25:100454.

23. Gentile P, Sterodimas A. Adipose stem cells (ASCs) and stromal vascularfraction (SVF) as a potential therapy in combating (COVID-19)-disease. AgingDis. 11: Copyright: © 2020 Gentile et al.; 2020. p. 465–469.

24. Niada S, Giannasi C, Gualerzi A, Banfi G, Brini AT. Differential proteomicanalysis predicts appropriate applications for the secretome of adipose-derived mesenchymal stem/stromal cells and dermal fibroblasts. Stem CellsInt. 2018;2018:7309031.

25. Gualerzi A, Niada S, Giannasi C, Picciolini S, Morasso C, Vanna R, et al. Ramanspectroscopy uncovers biochemical tissue-related features of extracellularvesicles from mesenchymal stromal cells. Sci Rep. 2017;7(1):9820.

26. Gualerzi A, Kooijmans SAA, Niada S, Picciolini S, Brini AT, Camussi G, et al.Raman spectroscopy as a quick tool to assess purity of extracellular vesiclepreparations and predict their functionality. J Extracell Vesicles. 2019;8(1):1568780.

27. Niada S, Giannasi C, Gomarasca M, Stanco D, Casati S, Brini AT. Adipose-derived stromal cell secretome reduces TNFalpha-induced hypertrophy andcatabolic markers in primary human articular chondrocytes. Stem Cell Res.2019;38:101463.

28. Laganà M, Arrigoni C, Lopa S, Sansone V, Zagra L, Moretti M, et al.Characterization of articular chondrocytes isolated from 211 osteoarthriticpatients. Cell Tissue Bank. 2014;15(1):59–66.

Giannasi et al. Stem Cell Research & Therapy (2020) 11:521 Page 13 of 15

29. de Girolamo L, Lopa S, Arrigoni E, Sartori MF, Baruffaldi Preis FW, Brini AT.Human adipose-derived stem cells isolated from young and elderly women:their differentiation potential and scaffold interaction during in vitroosteoblastic differentiation. Cytotherapy. 2009;11(6):793–803.

30. Jakob M, Démarteau O, Schäfer D, Stumm M, Heberer M, Martin I.Enzymatic digestion of adult human articular cartilage yields a smallfraction of the total available cells. Connect Tissue Res. 2003;44(3–4):173–80.

31. Giannasi C, Pagni G, Polenghi C, Niada S, Manfredi B, Brini AT, et al.Impact of dental implant surface modifications on adhesion andproliferation of primary human gingival keratinocytes and progenitorcells. Int J Periodontics Restorative Dent.2018;38(1):127–35.

32. Brini AT, Cocce V, Ferreira LM, Giannasi C, Cossellu G, Gianni AB, et al. Cell-mediated drug delivery by gingival interdental papilla mesenchymalstromal cells (GinPa-MSCs) loaded with paclitaxel. Expert Opin Drug Deliv.2016;13(6):789–98.

33. Niada S, Giannasi C, Ferreira LM, Milani A, Arrigoni E, Brini AT. 17beta-estradiol differently affects osteogenic differentiation of mesenchymal stem/stromal cells from adipose tissue and bone marrow. Differentiation. 2016;92(5):291–7.

34. Niada S, Ferreira LM, Arrigoni E, Addis A, Campagnol M, Broccaioli E, et al.Porcine adipose-derived stem cells from buccal fat pad and subcutaneousadipose tissue for future preclinical studies in oral surgery. Stem Cell ResTher. 2013;4(6):148.

35. Petrenko Y, Vackova I, Kekulova K, Chudickova M, Koci Z, Turnovcova K,et al. A comparative analysis of multipotent mesenchymal stromal cellsderived from different sources, with a focus on neuroregenerative potential.Sci Rep. 2020;10(1):4290.

36. Brini AT, Amodeo G, Ferreira LM, Milani A, Niada S, Moschetti G, et al.Therapeutic effect of human adipose-derived stem cells and theirsecretome in experimental diabetic pain. Sci Rep. 2017;7(1):9904.

37. Pergoli L, Cantone L, Favero C, Angelici L, Iodice S, Pinatel E, et al.Extracellular vesicle-packaged miRNA release after short-term exposure toparticulate matter is associated with increased coagulation. Part FibreToxicol. 2017;14(1):32.

38. Rezaie J, Nejati V, Khaksar M, Oryan A, Aghamohamadzadeh N, ShariatzadehMA, et al. Diabetic sera disrupted the normal exosome signaling pathway inhuman mesenchymal stem cells in vitro. Cell Tissue Res. 2018;374(3):555–65.

39. Ragni E, Perucca Orfei C, De Luca P, Lugano G, Vigano M, Colombini A,et al. Interaction with hyaluronan matrix and miRNA cargo as contributorsfor in vitro potential of mesenchymal stem cell-derived extracellular vesiclesin a model of human osteoarthritic synoviocytes. Stem Cell Res Ther. 2019;10(1):109.

40. Lötvall J, Hill AF, Hochberg F, Buzás EI, Di Vizio D, Gardiner C, et al. Minimalexperimental requirements for definition of extracellular vesicles and theirfunctions: a position statement from the International Society forExtracellular Vesicles. J Extracell Vesicles. 2014;3:26913.

41. Cocce V, Balducci L, Falchetti ML, Pascucci L, Ciusani E, Brini AT, et al.Fluorescent immortalized human adipose derived stromal cells (hASCs-TS/GFP+) for studying cell drug delivery mediated by microvesicles. AntiCancer Agents Med Chem. 2017;17(11):1578–85.

42. Caron MM, Emans PJ, Coolsen MM, Voss L, Surtel DA, Cremers A, et al.Redifferentiation of dedifferentiated human articular chondrocytes:comparison of 2D and 3D cultures. Osteoarthr Cartil. 2012;20(10):1170–8.

43. Goldring MB. Human chondrocyte cultures as models of cartilage-specificgene regulation. Methods Mol Med. 2005;107:69–95.

44. Westacott CI, Barakat AF, Wood L, Perry MJ, Neison P, Bisbinas I, et al. Tumornecrosis factor alpha can contribute to focal loss of cartilage inosteoarthritis. Osteoarthr Cartil. 2000;8(3):213–21.

45. Wisniewski JR, Zougman A, Nagaraj N, Mann M. Universal samplepreparation method for proteome analysis. Nat Methods. 2009;6(5):359–62.

46. Cox J, Neuhauser N, Michalski A, Scheltema RA, Olsen JV, Mann M.Andromeda: a peptide search engine integrated into the MaxQuantenvironment. J Proteome Res. 2011;10(4):1794–805.

47. Giannasi C, Niada S, Farronato D, Lombardi G, Manfredi B, Farronato G, et al.Nitrogen containing bisphosphonates impair the release of bonehomeostasis mediators and matrix production by human primary pre-osteoblasts. Int J Med Sci. 2019;16(1):23–32.

48. Singh P, Marcu KB, Goldring MB, Otero M. Phenotypic instability ofchondrocytes in osteoarthritis: on a path to hypertrophy. Ann N Y Acad Sci.2019;1442(1):17-34. https://doi.org/10.1111/nyas.13930.

49. Wang M, Sampson ER, Jin H, Li J, Ke QH, Im HJ, et al. MMP13 is a criticaltarget gene during the progression of osteoarthritis. Arthritis Res Ther. 2013;15(1):R5.

50. van der Kraan PM, van den Berg WB. Chondrocyte hypertrophy andosteoarthritis: role in initiation and progression of cartilage degeneration?Osteoarthr Cartil. 2012;20(3):223–32.

51. Li TF, Zuscik MJ, Ionescu AM, Zhang X, Rosier RN, Schwarz EM, et al. PGE2inhibits chondrocyte differentiation through PKA and PKC signaling. ExpCell Res. 2004;300(1):159–69.

52. Attur M, Al-Mussawir HE, Patel J, Kitay A, Dave M, Palmer G, et al.Prostaglandin E2 exerts catabolic effects in osteoarthritis cartilage: evidencefor signaling via the EP4 receptor. J Immunol. 2008;181(7):5082–8.

53. Donahue HJ, Qu RW, Genetos DC. Joint diseases: from connexins to gapjunctions. Nat Rev Rheumatol. 2017;14(1):42–51.

54. Gago-Fuentes R, Bechberger JF, Varela-Eirin M, Varela-Vazquez A, Acea B,Fonseca E, et al. The C-terminal domain of connexin43 modulates cartilagestructure via chondrocyte phenotypic changes. Oncotarget. 2016;7(45):73055–67.

55. Caplan AI. Mesenchymal stem cells: time to change the name! Stem CellsTransl Med. 2017;6(6):1445–51.

56. Bogatcheva NV, Coleman ME. Conditioned medium of mesenchymalstromal cells: a new class of therapeutics. Biochemistry (Mosc). 2019;84(11):1375–89.

57. Gentile P, Garcovich S. Advances in Regenerative Stem Cell Therapy inAndrogenic Alopecia and Hair Loss: Wnt pathway, Growth-Factor, andMesenchymal Stem Cell Signaling Impact Analysis on Cell Growth and HairFollicle Development. Cells. 2019;8(5):466. https://doi.org/10.3390/cells8050466.

58. Zhou BR, Xu Y, Guo SL, Wang Y, Zhu F, Permatasari F, et al. The effect ofconditioned media of adipose-derived stem cells on wound healing afterablative fractional carbon dioxide laser resurfacing. Biomed Res Int. 2013;2013:519126.

59. Fukuoka H, Narita K, Suga H. Hair regeneration therapy: application ofadipose-derived stem cells. Curr Stem Cell Res Ther. 2017;12(7):531–4.

60. Shin H, Won CH, Chung WK, Park BS. Up-to-date clinical trials of hairregeneration using conditioned media of adipose-derived stem cells inmale and female pattern hair loss. Curr Stem Cell Res Ther. 2017;12(7):524–30.

61. Katagiri W, Watanabe J, Toyama N, Osugi M, Sakaguchi K, Hibi H.Clinical study of bone regeneration by conditioned medium frommesenchymal stem cells after maxillary sinus floor elevation. ImplantDent. 2017;26(4):607–12.

62. Dahbour S, Jamali F, Alhattab D, Al-Radaideh A, Ababneh O, Al-RyalatN, et al. Mesenchymal stem cells and conditioned media in thetreatment of multiple sclerosis patients: clinical, ophthalmological andradiological assessments of safety and efficacy. CNS Neurosci Ther.2017;23(11):866–74.

63. Kordelas L, Rebmann V, Ludwig AK, Radtke S, Ruesing J, Doeppner TR, et al.MSC-derived exosomes: a novel tool to treat therapy-refractory graft-versus-host disease. Leukemia. 2014;28:970–3.

64. Nassar W, El-Ansary M, Sabry D, Mostafa MA, Fayad T, Kotb E, et al. Umbilicalcord mesenchymal stem cells derived extracellular vesicles can safelyameliorate the progression of chronic kidney diseases. Biomater Res. 2016;20:21.

65. Fujita Y, Kadota T, Araya J, Ochiya T, Kuwano K. Clinical Application ofMesenchymal Stem Cell-Derived Extracellular Vesicle-Based Therapeutics forInflammatory Lung Diseases. J Clin Med. 2018;7(10):355. https://doi.org/10.3390/jcm7100355.

66. Mitchell R, Mellows B, Sheard J, Antonioli M, Kretz O, Chambers D, et al.Secretome of adipose-derived mesenchymal stem cells promotes skeletalmuscle regeneration through synergistic action of extracellular vesicle cargoand soluble proteins. Stem Cell Res Ther. 2019;10(1):116.

67. Drommelschmidt K, Serdar M, Bendix I, Herz J, Bertling F, Prager S, et al.Mesenchymal stem cell-derived extracellular vesicles ameliorateinflammation-induced preterm brain injury. Brain Behav Immun. 2017;60:220–32.

Giannasi et al. Stem Cell Research & Therapy (2020) 11:521 Page 14 of 15

68. Vergauwen G, Dhondt B, Van Deun J, De Smedt E, Berx G, Timmerman E,et al. Confounding factors of ultrafiltration and protein analysis inextracellular vesicle research. Sci Rep. 2017;7(1):2704.

69. Brennan K, Martin K, FitzGerald SP, O'Sullivan J, Wu Y, Blanco A, RichardsonC, Mc Gee MM. A comparison of methods for the isolation and separationof extracellular vesicles from protein and lipid particles in human serum. SciRep. 2020;10(1):1039. https://doi.org/10.1038/s41598-020-57497-7.

70. Takov K, Yellon DM, Davidson SM. Comparison of small extracellular vesiclesisolated from plasma by ultracentrifugation or size-exclusionchromatography: yield, purity and functional potential. J Extracell Vesicles.2019;8(1):1560809.

71. Tang YT, Huang YY, Zheng L, Qin SH, Xu XP, An TX, et al. Comparison ofisolation methods of exosomes and exosomal RNA from cell culturemedium and serum. Int J Mol Med. 2017;40(3):834–44.

72. Liu J, Khalil RA. Matrix metalloproteinase inhibitors as investigational andtherapeutic tools in unrestrained tissue remodeling and pathologicaldisorders. Prog Mol Biol Transl Sci. 2017;148:355–420.

73. Hu J, Van den Steen PE, Sang QX, Opdenakker G. Matrix metalloproteinaseinhibitors as therapy for inflammatory and vascular diseases. Nat Rev DrugDiscov. 2007;6(6):480–98.

74. Kawasaki Y, Xu ZZ, Wang X, Park JY, Zhuang ZY, Tan PH, et al. Distinct rolesof matrix metalloproteases in the early- and late-phase development ofneuropathic pain. Nat Med. 2008;14(3):331–6.

75. Tchetina EV, Di Battista JA, Zukor DJ, Antoniou J, Poole AR. ProstaglandinPGE2 at very low concentrations suppresses collagen cleavage in culturedhuman osteoarthritic articular cartilage: this involves a decrease inexpression of proinflammatory genes, collagenases and COL10A1, a genelinked to chondrocyte hypertrophy. Arthritis Res Ther. 2007;9(4):R75.

76. Li X, Ellman M, Muddasani P, Wang JH, Cs-Szabo G, van Wijnen AJ, et al.Prostaglandin E2 and its cognate EP receptors control human adult articularcartilage homeostasis and are linked to the pathophysiology ofosteoarthritis. Arthritis Rheum. 2009;60(2):513–23.

77. Vonk LA, van Dooremalen SFJ, Liv N, Klumperman J, Coffer PJ, Saris DBF,et al. Mesenchymal stromal/stem cell-derived extracellular vesicles promotehuman cartilage regeneration in vitro. Theranostics. 2018;8(4):906–20.

78. Roman-Blas JA, Jimenez SA. NF-kappaB as a potential therapeutic target inosteoarthritis and rheumatoid arthritis. Osteoarthr Cartil. 2006;14(9):839–48.

79. Gomzikova M, Kletukhina S, Kurbangaleeva S, Rizvanov A. Evaluation ofcytochalasin B-induced membrane vesicles fusion specificity with targetcells. Biomed Res Int. 2018;2018:7053623.

80. Gu J, Lu Y, Li F, Qiao L, Wang Q, Li N, et al. Identification andcharacterization of the novel Col10a1 regulatory mechanism duringchondrocyte hypertrophic differentiation. Cell Death Dis. 2014;5(10):e1469.

81. Vonk LA, Kragten AH, Dhert WJ, Saris DB, Creemers LB. Overexpression ofhsa-miR-148a promotes cartilage production and inhibits cartilagedegradation by osteoarthritic chondrocytes. Osteoarthr Cartil. 2014;22(1):145–53.

82. Ragni E, Perucca Orfei C, De Luca P, Mondadori C, Viganò M, Colombini A,et al. Inflammatory priming enhances mesenchymal stromal cell secretomepotential as a clinical product for regenerative medicine approachesthrough secreted factors and EV-miRNAs: the example of joint disease. StemCell Res Ther. 2020;11(1):165.

83. Choocheep K, Hatano S, Takagi H, Watanabe H, Kimata K, Kongtawelert P.Versican facilitates chondrocyte differentiation and regulates jointmorphogenesis. J Biol Chem. 2010;285(27):21114–25.

84. Zhang FF, Morioka N, Kitamura T, Hisaoka-Nakashima K, Nakata Y.Proinflammatory cytokines downregulate connexin 43-gap junctions via theubiquitin-proteasome system in rat spinal astrocytes. Biochem Biophys ResCommun. 2015;464(4):1202–8.

85. Kimura K, Orita T, Morishige N, Nishida T, Sonoda KH. Role of the JNKsignaling pathway in downregulation of connexin43 by TNF-alpha inhuman corneal fibroblasts. Curr Eye Res. 2013;38(9):926–32.

86. De Bock M, Wang N, Decrock E, Bultynck G, Leybaert L. Intracellularcleavage of the Cx43 C-terminal domain by matrix-metalloproteases: anovel contributor to inflammation? Mediat Inflamm. 2015;2015:257471.

87. Embree MC, Kilts TM, Ono M, Inkson CA, Syed-Picard F, Karsdal MA, et al.Biglycan and fibromodulin have essential roles in regulatingchondrogenesis and extracellular matrix turnover in temporomandibularjoint osteoarthritis. Am J Pathol. 2010;176(2):812–26.

88. Kessler E, Takahara K, Biniaminov L, Brusel M, Greenspan DS. Bonemorphogenetic protein-1: the type I procollagen C-proteinase. Science.1996;271(5247):360–2.

89. Hopkins DR, Keles S, Greenspan DS. The bone morphogenetic protein 1/Tolloid-like metalloproteinases. Matrix Biol. 2007;26(7):508–23.

90. Snelling SJ, Davidson RK, Swingler TE, Le LT, Barter MJ, Culley KL, et al.Dickkopf-3 is upregulated in osteoarthritis and has a chondroprotective role.Osteoarthr Cartil. 2016;24(5):883–91.

91. Wen CY, Wu CB, Tang B, Wang T, Yan CH, Lu WW, et al. Collagen fibrilstiffening in osteoarthritic cartilage of human beings revealed by atomicforce microscopy. Osteoarthr Cartil. 2012;20(8):916–22.

92. Shen J, Li S, Chen D. TGF-β signaling and the development of osteoarthritis.Bone Res. 2014;2:14002–09. https://doi.org/10.1038/boneres.2014.2.

93. Djouad F, Delorme B, Maurice M, Bony C, Apparailly F, Louis-Plence P, et al.Microenvironmental changes during differentiation of mesenchymal stemcells towards chondrocytes. Arthritis Res Ther. 2007;9(2):R33.

94. Yang CY, Chanalaris A, Troeberg L. ADAMTS and ADAM metalloproteinasesin osteoarthritis - looking beyond the ‘usual suspects’. Osteoarthr Cartil.2017;25(7):1000–9.

95. Scheller J, Chalaris A, Garbers C, Rose-John S. ADAM17: a molecular switchto control inflammation and tissue regeneration. Trends Immunol. 2011;32(8):380–7.

Publisher’s NoteSpringer Nature remains neutral with regard to jurisdictional claims inpublished maps and institutional affiliations.

Giannasi et al. Stem Cell Research & Therapy (2020) 11:521 Page 15 of 15